CHFD 5110 Exam 1

• Concerned with developing, exploring, or testing theories
• Needed to tie our ideas to a bigger body of research

Based on direct observations and measurements of reality

Inferences based on probabilities

Chance of something happening

Interested in what leads to what (cause & effect)

• Designed to document what's going on or what exists
• What is the opinion/attitude of a group of people?
• Ex: What are the traits of a happy marriage?

• Looks at the relationships between one or two variables
• How is the person's opinion/attitude related to other factors?
• Ex: gender and voting preferences

• What factors affect ("cause") their opinions?
• Ex: How do you find out if x drug works?

• Takes place at a single point in time
• Pro's: Fast, less expensive, less dropouts
• Con's: Not comparing apples to apples
• Ex: Contact ppl at diff. ages to ask about marriage

• A study that takes place over time
• Measurement occurs on at least 2 separate occasions
• Pro's: Comparing the same ppl
• Con's: Takes long, more expensive, people can drop out
• Ex: Keep track of the same ppl over time to see how their marriage goes

• Any observation that can take different values
• Ex: Sex/gender, Height
• Something in the population that you want to study

• A specific value on the variable
• Ex: If the variable is gender, the attribute is male or female

Causation

• Leads to or causes something else
• Ex: Gender (leading to the cause of the views of the president)

• Is affected by other variable(s)
• Ex: Views on the president (affected by gender)

• 
• Ex: Your lifeline and GPA

• 
• As one goes up, the other does too
• Ex: Your salary and your level of education

• 
• As one goes down, the other goes up
• Ex: Paranoia level & self-esteem

• 
• Ex: Time spent with your partner & marital satisfaction

A specific statement of prediction

• The alternative is the one you want to support
• Don't prove your hypothesis, just disprove the opposite

• If you discredit the left side, you prove the right side
• 

• Don't know what it will do, so you test both sides
• Any change against no change
• 

Anything you can put a number on

Anything else (besides what you can put a number on)

• Works to confirm to disconfirm a theory**
• Helps to provide assumptions
• Gives us a guide
• Helps us talk about it
• Theory -> Hypothesis -> Observation -> Confirmation
• Ex: Drug trials, Are they keeping off the weight they lost?

• The idea is to create a theory
• Building understanding**
• Observation -> Pattern -> Tentative Hypothesis -> Theory
• Ex: What's it like to...? What are the effects of...?

An error in reasoning, usually based on mistaken assumptions

• Mistaken conclusion about an individual based on an analysis of a group
• Ex: All babies start to walk between 9-12 months and if mine doesn't it's a huge problem

• Reaching a group conclusion on the basic of exceptional cases (ie- racism)
• "I see ___ happen, so everyone that's like that does this"
• Ex: My child walked before they crawled, so all children should walk before they crawl
• Another Ex: That woman is a bad driver, so all women are bad drivers

• The gap between Truth and truth
• Conclustions reached about the quality of different parts of the research methodology

• The degree to which the conclusions in your study would hold for other persons in other places and at other times
• Can you generalize the results? Who does this apply to?
• Threats: People, places, time
• Problem: By the time you get to the sample, it's not very random
• Two approaches: Sampling model & Proximal similarity model

• Focus on the links
• Take a big idea and make it measureable
• Ex: Do a questionnaire to measure trust v. mistrust

• Is there a relationship between the 2 variables?
• Ex: Is there a relationship between using the workbook and course knowledge?

• Assuming there's a relationship, is it a causal one?
• Just because there's a relationship doesn't mean it's a causal one

Not forced to participate

The know exactly what's going to happen (risks & benefits) before participation

• The researcher can identify the participants but isn't going to share
• Most studies are confidential (as opposed to anonymous)

The researcher cannot identify the participants and cannot link their response back to them

• Have to provide service if something is working/not working
• Ex: Find that one group is doing really well with this treatment, so you have to give it to the other group too

Makes sure you're being respectful and honouring peoples' rights

• Consent form makrs a point of change
• Indicates that ethical concern goes hand-in-hand with research on humans

• A guide, not a law
• Increased protection of human subjects
• Voluntary, informed consent
• Animals must be tested before humans
• Only scientific professionals can conduct research
• Subject can withdraw at any time
• Researchers should be prepared to stop at any time

• Risks reasonable in relation to benefits
• Risks kept at the minimum necessary
• Equitable selection
• Informed consent
• Privacy & confidentiality
• Continuous monitoring of data

Model for generalizing in which you identify your population, draw a fair sample, conduct your research, and then generalize your population to other groups

Model for generalizing from your study to another context based on the degree to which the other context is similar to yours

The group to whom you wish to generalize

• Who do you want to generalize to?
• Not always easy to get
• Ex: People who live in Athens

• Who can you get access to?
• Ex: You don't have access to every single person who lives in Athens, but you do have access to the people in the census, those who are registered at the DMV, etc.

The listing of accessible people in your population

Who's in your study

• The results of the study
• Ex: The average self-esteem of so-and-so-group is 3.72

• What the number actually is in the population
• Ex: Study says the average self-esteem is 3.72 but it's actually 3.91
• Usually theoretical unless it's a very narrow population
• Calculated based on probability theory

• How you get from a sample statistic to an estimate of the population parameter
• The distribution of an infinite number of samples of the same size as the sample of your study

• Utilizes some form of random selection
• All units in the population have an equal probability of being chosen

• Probability sampling
• Everyone has an equal chance of being chosen
• Use a table of random numbers generated by computer
• Use anytime*
• Pros: Easy to do, Easy to explain*
• Cons: Requires a sample list from which to select, Generalizable, but may not be representative of subgroups (could leave someone out in your sample)*

• Probability sampling
• Divide your population into subgroups that are homogenous and then use a simple random sample for each subgroup
• Use: When concerned ab. underrepresenting smaller groups*
• Pros: Allows you to oversample minority groups to assure enough for subgroup analyses*
• Cons: Requires a sample list from which to select, Doesn't generalize as well as simple random sampling*

• Probability sampling
• If you have a list of 100 people and want 20, you roll a dice and roll a 2. You start with the second person and then every 5th person down*
• Pros: Don't have to cont thru all the elements in the list to find the ones randomly selected*
• Cons: If the order of elements is nonrandom, could be systematic bias*

• Do a simple random sample of clusters & then randomly select clusters. You need to measure everyone within a sampled cluster
• Usually geographical, but doesn't have to be*
• Pros: More efficient when surveying a geographical area*
• Cons: Usually not used alone*

• Probability sampling
• Combine any of the sampling techniques to create a more efficient or more effective sample than the use of any one sampling type can achieve on its own
• Use: Anytime*
• Pros: Combines sophistication with efficiency*
• Cons: Can be complex and hard to explain*

• Doesn't involve random selection
• May be representative but can't depend on the rationale of probability theory
• Used when it's not feasible, practical, or theoretically sensible to do random sampling (don't want to randomly select a person who doesn't apply)

• Nonprobability sampling
• One of the most common methods of sampling
• "Man on the street"
• Volunteers or subjects who are "conveniently" available" (like college students)
• Use: Anytime*
• Pros: Very easy*
• Cons: Very hard to generalize, Likely to be biased*

• Nonprobability sampling
• Sampling with a "purpose" in mind
• Useful in reaching a targeted sample quickly
• Target population is reached with over-representation of subgroups that are more readily accessible

• Nonprobability sampling - Purposive
• Sampling the most frequent or typical case
• Difficult to determine what a "typical case" is
• Useful for informal sampling contexts
• Use: When you only want to measure a typical respondent*
• Pros: Easily understood*
• Cons: Generalizeable

• Nonprobability sampling - Purposive
• Assembling of a sample of persons with known or demonstrable expertise in some area
• "Panel of experts"
• May be useful for providing evidence as to the validity of another sampling approach you have chosen
• Use: With other sampling strategies*
• Pros: Experts opinions support research conclusions*
• Cons: Likely to be biased, Limited external validity

• Nonprobability sampling - Purposive
• Sample selected nonrandomly according to some fixed quota
• Proportional quota sampling used to represent the major characteristics of the population of interest by sampling a proportional amount of each
• Nonproportional quota sampling used to supply a minimum number of units in each category but not concerned with proportions
• Use: When you want to represent subgroups*
• Pros: Broader sample, Will match what you want it to, There will never be outliers*
• Cons: You could potentially not get the ppl you're looking for, Limits your range (could be missing info that could be valuable), Allows you to be biased*

• Nonprobability sampling - Purposive
• Used to provide a sample that will include "all" the views or opinions without regard to proportional representation
• Use: Sampling for diversity*
• Pros: Easy to use and explain, Better when you're interested in samily for variety*
• Cons: Won't represent population views proportionately*

• Ppl meeting the criteria for inclusion in the sample are identified and then they recommend others they know who meet the criteria
• Use: With hard-to-reach populations*
• Pros: Can be used when there is no sampling frame*
• Cons: Low external validity*